Confronting Computer Models of Children’s Word Problem Solving With Empirical Data

Abstract

Over the past few years a considerable number of studies have been undertaken that try to assess and to get a better understanding of young children’s problem solving with respect to simple addition and subtraction word problems (for example: “Pete has 8 marbles; Tom gives him 5 more marbles; how many marbles does Pete have now?”). This research has resulted in the construction of theoretical models of children’s problem-solving processes. Some of these models have been translated into computer programs (Riley, Greeno and Heller, 1983; Briars and Larkin, 1984). These computer models simulate the knowledge and actions that are required to understand and solve addition and subtraction word problems. On the basis of such models predictions can be made concerning the level of difficulty of different types of word problems, the strategies applied to solve them, and the errors that one can expect as a consequence of specific shortcomings in children’s knowledge and actions. Using qualitative material collected by self-reporting techniques, such as thinking aloud and retrospection and by systematic behaviour observation, one can verify, to what degree the computer models fit empirical data on children’s problem solving. In the present contribution this will be done with respect to the theoretical analysis with computer programs developed by (1983).